1/************************************************* 2* PCRE testing program * 3*************************************************/ 4 5/* This program was hacked up as a tester for PCRE. I really should have 6written it more tidily in the first place. Will I ever learn? It has grown and 7been extended and consequently is now rather, er, *very* untidy in places. 8 9----------------------------------------------------------------------------- 10Redistribution and use in source and binary forms, with or without 11modification, are permitted provided that the following conditions are met: 12 13 * Redistributions of source code must retain the above copyright notice, 14 this list of conditions and the following disclaimer. 15 16 * Redistributions in binary form must reproduce the above copyright 17 notice, this list of conditions and the following disclaimer in the 18 documentation and/or other materials provided with the distribution. 19 20 * Neither the name of the University of Cambridge nor the names of its 21 contributors may be used to endorse or promote products derived from 22 this software without specific prior written permission. 23 24THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" 25AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 26IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 27ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE 28LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 29CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 30SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 31INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 32CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 33ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 34POSSIBILITY OF SUCH DAMAGE. 35----------------------------------------------------------------------------- 36*/ 37 38 39#ifdef HAVE_CONFIG_H 40#include "config.h" 41#endif 42 43#include <ctype.h> 44#include <stdio.h> 45#include <string.h> 46#include <stdlib.h> 47#include <time.h> 48#include <locale.h> 49#include <errno.h> 50 51#ifdef SUPPORT_LIBREADLINE 52#ifdef HAVE_UNISTD_H 53#include <unistd.h> 54#endif 55#include <readline/readline.h> 56#include <readline/history.h> 57#endif 58 59 60/* A number of things vary for Windows builds. Originally, pcretest opened its 61input and output without "b"; then I was told that "b" was needed in some 62environments, so it was added for release 5.0 to both the input and output. (It 63makes no difference on Unix-like systems.) Later I was told that it is wrong 64for the input on Windows. I've now abstracted the modes into two macros that 65are set here, to make it easier to fiddle with them, and removed "b" from the 66input mode under Windows. */ 67 68#if defined(_WIN32) || defined(WIN32) 69#include <io.h> /* For _setmode() */ 70#include <fcntl.h> /* For _O_BINARY */ 71#define INPUT_MODE "r" 72#define OUTPUT_MODE "wb" 73 74#ifndef isatty 75#define isatty _isatty /* This is what Windows calls them, I'm told, */ 76#endif /* though in some environments they seem to */ 77 /* be already defined, hence the #ifndefs. */ 78#ifndef fileno 79#define fileno _fileno 80#endif 81 82#else 83#include <sys/time.h> /* These two includes are needed */ 84#include <sys/resource.h> /* for setrlimit(). */ 85#define INPUT_MODE "rb" 86#define OUTPUT_MODE "wb" 87#endif 88 89 90/* We have to include pcre_internal.h because we need the internal info for 91displaying the results of pcre_study() and we also need to know about the 92internal macros, structures, and other internal data values; pcretest has 93"inside information" compared to a program that strictly follows the PCRE API. 94 95Although pcre_internal.h does itself include pcre.h, we explicitly include it 96here before pcre_internal.h so that the PCRE_EXP_xxx macros get set 97appropriately for an application, not for building PCRE. */ 98 99#include "pcre.h" 100#include "pcre_internal.h" 101 102/* We need access to some of the data tables that PCRE uses. So as not to have 103to keep two copies, we include the source file here, changing the names of the 104external symbols to prevent clashes. */ 105 106#define _pcre_ucp_gentype ucp_gentype 107#define _pcre_utf8_table1 utf8_table1 108#define _pcre_utf8_table1_size utf8_table1_size 109#define _pcre_utf8_table2 utf8_table2 110#define _pcre_utf8_table3 utf8_table3 111#define _pcre_utf8_table4 utf8_table4 112#define _pcre_utt utt 113#define _pcre_utt_size utt_size 114#define _pcre_utt_names utt_names 115#define _pcre_OP_lengths OP_lengths 116 117#include "pcre_tables.c" 118 119/* We also need the pcre_printint() function for printing out compiled 120patterns. This function is in a separate file so that it can be included in 121pcre_compile.c when that module is compiled with debugging enabled. It needs to 122know which case is being compiled. */ 123 124#define COMPILING_PCRETEST 125#include "pcre_printint.src" 126 127/* The definition of the macro PRINTABLE, which determines whether to print an 128output character as-is or as a hex value when showing compiled patterns, is 129contained in the printint.src file. We uses it here also, in cases when the 130locale has not been explicitly changed, so as to get consistent output from 131systems that differ in their output from isprint() even in the "C" locale. */ 132 133#define PRINTHEX(c) (locale_set? isprint(c) : PRINTABLE(c)) 134 135/* It is possible to compile this test program without including support for 136testing the POSIX interface, though this is not available via the standard 137Makefile. */ 138 139#if !defined NOPOSIX 140#include "pcreposix.h" 141#endif 142 143/* It is also possible, for the benefit of the version currently imported into 144Exim, to build pcretest without support for UTF8 (define NOUTF8), without the 145interface to the DFA matcher (NODFA), and without the doublecheck of the old 146"info" function (define NOINFOCHECK). In fact, we automatically cut out the 147UTF8 support if PCRE is built without it. */ 148 149#ifndef SUPPORT_UTF8 150#ifndef NOUTF8 151#define NOUTF8 152#endif 153#endif 154 155 156/* Other parameters */ 157 158#ifndef CLOCKS_PER_SEC 159#ifdef CLK_TCK 160#define CLOCKS_PER_SEC CLK_TCK 161#else 162#define CLOCKS_PER_SEC 100 163#endif 164#endif 165 166/* This is the default loop count for timing. */ 167 168#define LOOPREPEAT 500000 169 170/* Static variables */ 171 172static FILE *outfile; 173static int log_store = 0; 174static int callout_count; 175static int callout_extra; 176static int callout_fail_count; 177static int callout_fail_id; 178static int debug_lengths; 179static int first_callout; 180static int locale_set = 0; 181static int show_malloc; 182static int use_utf8; 183static size_t gotten_store; 184 185/* The buffers grow automatically if very long input lines are encountered. */ 186 187static int buffer_size = 50000; 188static uschar *buffer = NULL; 189static uschar *dbuffer = NULL; 190static uschar *pbuffer = NULL; 191 192 193 194/************************************************* 195* Read or extend an input line * 196*************************************************/ 197 198/* Input lines are read into buffer, but both patterns and data lines can be 199continued over multiple input lines. In addition, if the buffer fills up, we 200want to automatically expand it so as to be able to handle extremely large 201lines that are needed for certain stress tests. When the input buffer is 202expanded, the other two buffers must also be expanded likewise, and the 203contents of pbuffer, which are a copy of the input for callouts, must be 204preserved (for when expansion happens for a data line). This is not the most 205optimal way of handling this, but hey, this is just a test program! 206 207Arguments: 208 f the file to read 209 start where in buffer to start (this *must* be within buffer) 210 prompt for stdin or readline() 211 212Returns: pointer to the start of new data 213 could be a copy of start, or could be moved 214 NULL if no data read and EOF reached 215*/ 216 217static uschar * 218extend_inputline(FILE *f, uschar *start, const char *prompt) 219{ 220uschar *here = start; 221 222for (;;) 223 { 224 int rlen = buffer_size - (here - buffer); 225 226 if (rlen > 1000) 227 { 228 int dlen; 229 230 /* If libreadline support is required, use readline() to read a line if the 231 input is a terminal. Note that readline() removes the trailing newline, so 232 we must put it back again, to be compatible with fgets(). */ 233 234#ifdef SUPPORT_LIBREADLINE 235 if (isatty(fileno(f))) 236 { 237 size_t len; 238 char *s = readline(prompt); 239 if (s == NULL) return (here == start)? NULL : start; 240 len = strlen(s); 241 if (len > 0) add_history(s); 242 if (len > rlen - 1) len = rlen - 1; 243 memcpy(here, s, len); 244 here[len] = '\n'; 245 here[len+1] = 0; 246 free(s); 247 } 248 else 249#endif 250 251 /* Read the next line by normal means, prompting if the file is stdin. */ 252 253 { 254 if (f == stdin) printf(prompt); 255 if (fgets((char *)here, rlen, f) == NULL) 256 return (here == start)? NULL : start; 257 } 258 259 dlen = (int)strlen((char *)here); 260 if (dlen > 0 && here[dlen - 1] == '\n') return start; 261 here += dlen; 262 } 263 264 else 265 { 266 int new_buffer_size = 2*buffer_size; 267 uschar *new_buffer = (unsigned char *)malloc(new_buffer_size); 268 uschar *new_dbuffer = (unsigned char *)malloc(new_buffer_size); 269 uschar *new_pbuffer = (unsigned char *)malloc(new_buffer_size); 270 271 if (new_buffer == NULL || new_dbuffer == NULL || new_pbuffer == NULL) 272 { 273 fprintf(stderr, "pcretest: malloc(%d) failed\n", new_buffer_size); 274 exit(1); 275 } 276 277 memcpy(new_buffer, buffer, buffer_size); 278 memcpy(new_pbuffer, pbuffer, buffer_size); 279 280 buffer_size = new_buffer_size; 281 282 start = new_buffer + (start - buffer); 283 here = new_buffer + (here - buffer); 284 285 free(buffer); 286 free(dbuffer); 287 free(pbuffer); 288 289 buffer = new_buffer; 290 dbuffer = new_dbuffer; 291 pbuffer = new_pbuffer; 292 } 293 } 294 295return NULL; /* Control never gets here */ 296} 297 298 299 300 301 302 303 304/************************************************* 305* Read number from string * 306*************************************************/ 307 308/* We don't use strtoul() because SunOS4 doesn't have it. Rather than mess 309around with conditional compilation, just do the job by hand. It is only used 310for unpicking arguments, so just keep it simple. 311 312Arguments: 313 str string to be converted 314 endptr where to put the end pointer 315 316Returns: the unsigned long 317*/ 318 319static int 320get_value(unsigned char *str, unsigned char **endptr) 321{ 322int result = 0; 323while(*str != 0 && isspace(*str)) str++; 324while (isdigit(*str)) result = result * 10 + (int)(*str++ - '0'); 325*endptr = str; 326return(result); 327} 328 329 330 331 332/************************************************* 333* Convert UTF-8 string to value * 334*************************************************/ 335 336/* This function takes one or more bytes that represents a UTF-8 character, 337and returns the value of the character. 338 339Argument: 340 utf8bytes a pointer to the byte vector 341 vptr a pointer to an int to receive the value 342 343Returns: > 0 => the number of bytes consumed 344 -6 to 0 => malformed UTF-8 character at offset = (-return) 345*/ 346 347#if !defined NOUTF8 348 349static int 350utf82ord(unsigned char *utf8bytes, int *vptr) 351{ 352int c = *utf8bytes++; 353int d = c; 354int i, j, s; 355 356for (i = -1; i < 6; i++) /* i is number of additional bytes */ 357 { 358 if ((d & 0x80) == 0) break; 359 d <<= 1; 360 } 361 362if (i == -1) { *vptr = c; return 1; } /* ascii character */ 363if (i == 0 || i == 6) return 0; /* invalid UTF-8 */ 364 365/* i now has a value in the range 1-5 */ 366 367s = 6*i; 368d = (c & utf8_table3[i]) << s; 369 370for (j = 0; j < i; j++) 371 { 372 c = *utf8bytes++; 373 if ((c & 0xc0) != 0x80) return -(j+1); 374 s -= 6; 375 d |= (c & 0x3f) << s; 376 } 377 378/* Check that encoding was the correct unique one */ 379 380for (j = 0; j < utf8_table1_size; j++) 381 if (d <= utf8_table1[j]) break; 382if (j != i) return -(i+1); 383 384/* Valid value */ 385 386*vptr = d; 387return i+1; 388} 389 390#endif 391 392 393 394/************************************************* 395* Convert character value to UTF-8 * 396*************************************************/ 397 398/* This function takes an integer value in the range 0 - 0x7fffffff 399and encodes it as a UTF-8 character in 0 to 6 bytes. 400 401Arguments: 402 cvalue the character value 403 utf8bytes pointer to buffer for result - at least 6 bytes long 404 405Returns: number of characters placed in the buffer 406*/ 407 408#if !defined NOUTF8 409 410static int 411ord2utf8(int cvalue, uschar *utf8bytes) 412{ 413register int i, j; 414for (i = 0; i < utf8_table1_size; i++) 415 if (cvalue <= utf8_table1[i]) break; 416utf8bytes += i; 417for (j = i; j > 0; j--) 418 { 419 *utf8bytes-- = 0x80 | (cvalue & 0x3f); 420 cvalue >>= 6; 421 } 422*utf8bytes = utf8_table2[i] | cvalue; 423return i + 1; 424} 425 426#endif 427 428 429 430/************************************************* 431* Print character string * 432*************************************************/ 433 434/* Character string printing function. Must handle UTF-8 strings in utf8 435mode. Yields number of characters printed. If handed a NULL file, just counts 436chars without printing. */ 437 438static int pchars(unsigned char *p, int length, FILE *f) 439{ 440int c = 0; 441int yield = 0; 442 443while (length-- > 0) 444 { 445#if !defined NOUTF8 446 if (use_utf8) 447 { 448 int rc = utf82ord(p, &c); 449 450 if (rc > 0 && rc <= length + 1) /* Mustn't run over the end */ 451 { 452 length -= rc - 1; 453 p += rc; 454 if (PRINTHEX(c)) 455 { 456 if (f != NULL) fprintf(f, "%c", c); 457 yield++; 458 } 459 else 460 { 461 int n = 4; 462 if (f != NULL) fprintf(f, "\\x{%02x}", c); 463 yield += (n <= 0x000000ff)? 2 : 464 (n <= 0x00000fff)? 3 : 465 (n <= 0x0000ffff)? 4 : 466 (n <= 0x000fffff)? 5 : 6; 467 } 468 continue; 469 } 470 } 471#endif 472 473 /* Not UTF-8, or malformed UTF-8 */ 474 475 c = *p++; 476 if (PRINTHEX(c)) 477 { 478 if (f != NULL) fprintf(f, "%c", c); 479 yield++; 480 } 481 else 482 { 483 if (f != NULL) fprintf(f, "\\x%02x", c); 484 yield += 4; 485 } 486 } 487 488return yield; 489} 490 491 492 493/************************************************* 494* Callout function * 495*************************************************/ 496 497/* Called from PCRE as a result of the (?C) item. We print out where we are in 498the match. Yield zero unless more callouts than the fail count, or the callout 499data is not zero. */ 500 501static int callout(pcre_callout_block *cb) 502{ 503FILE *f = (first_callout | callout_extra)? outfile : NULL; 504int i, pre_start, post_start, subject_length; 505 506if (callout_extra) 507 { 508 fprintf(f, "Callout %d: last capture = %d\n", 509 cb->callout_number, cb->capture_last); 510 511 for (i = 0; i < cb->capture_top * 2; i += 2) 512 { 513 if (cb->offset_vector[i] < 0) 514 fprintf(f, "%2d: <unset>\n", i/2); 515 else 516 { 517 fprintf(f, "%2d: ", i/2); 518 (void)pchars((unsigned char *)cb->subject + cb->offset_vector[i], 519 cb->offset_vector[i+1] - cb->offset_vector[i], f); 520 fprintf(f, "\n"); 521 } 522 } 523 } 524 525/* Re-print the subject in canonical form, the first time or if giving full 526datails. On subsequent calls in the same match, we use pchars just to find the 527printed lengths of the substrings. */ 528 529if (f != NULL) fprintf(f, "--->"); 530 531pre_start = pchars((unsigned char *)cb->subject, cb->start_match, f); 532post_start = pchars((unsigned char *)(cb->subject + cb->start_match), 533 cb->current_position - cb->start_match, f); 534 535subject_length = pchars((unsigned char *)cb->subject, cb->subject_length, NULL); 536 537(void)pchars((unsigned char *)(cb->subject + cb->current_position), 538 cb->subject_length - cb->current_position, f); 539 540if (f != NULL) fprintf(f, "\n"); 541 542/* Always print appropriate indicators, with callout number if not already 543shown. For automatic callouts, show the pattern offset. */ 544 545if (cb->callout_number == 255) 546 { 547 fprintf(outfile, "%+3d ", cb->pattern_position); 548 if (cb->pattern_position > 99) fprintf(outfile, "\n "); 549 } 550else 551 { 552 if (callout_extra) fprintf(outfile, " "); 553 else fprintf(outfile, "%3d ", cb->callout_number); 554 } 555 556for (i = 0; i < pre_start; i++) fprintf(outfile, " "); 557fprintf(outfile, "^"); 558 559if (post_start > 0) 560 { 561 for (i = 0; i < post_start - 1; i++) fprintf(outfile, " "); 562 fprintf(outfile, "^"); 563 } 564 565for (i = 0; i < subject_length - pre_start - post_start + 4; i++) 566 fprintf(outfile, " "); 567 568fprintf(outfile, "%.*s", (cb->next_item_length == 0)? 1 : cb->next_item_length, 569 pbuffer + cb->pattern_position); 570 571fprintf(outfile, "\n"); 572first_callout = 0; 573 574if (cb->callout_data != NULL) 575 { 576 int callout_data = *((int *)(cb->callout_data)); 577 if (callout_data != 0) 578 { 579 fprintf(outfile, "Callout data = %d\n", callout_data); 580 return callout_data; 581 } 582 } 583 584return (cb->callout_number != callout_fail_id)? 0 : 585 (++callout_count >= callout_fail_count)? 1 : 0; 586} 587 588 589/************************************************* 590* Local malloc functions * 591*************************************************/ 592 593/* Alternative malloc function, to test functionality and show the size of the 594compiled re. */ 595 596static void *new_malloc(size_t size) 597{ 598void *block = malloc(size); 599gotten_store = size; 600if (show_malloc) 601 fprintf(outfile, "malloc %3d %p\n", (int)size, block); 602return block; 603} 604 605static void new_free(void *block) 606{ 607if (show_malloc) 608 fprintf(outfile, "free %p\n", block); 609free(block); 610} 611 612 613/* For recursion malloc/free, to test stacking calls */ 614 615static void *stack_malloc(size_t size) 616{ 617void *block = malloc(size); 618if (show_malloc) 619 fprintf(outfile, "stack_malloc %3d %p\n", (int)size, block); 620return block; 621} 622 623static void stack_free(void *block) 624{ 625if (show_malloc) 626 fprintf(outfile, "stack_free %p\n", block); 627free(block); 628} 629 630 631/************************************************* 632* Call pcre_fullinfo() * 633*************************************************/ 634 635/* Get one piece of information from the pcre_fullinfo() function */ 636 637static void new_info(pcre *re, pcre_extra *study, int option, void *ptr) 638{ 639int rc; 640if ((rc = pcre_fullinfo(re, study, option, ptr)) < 0) 641 fprintf(outfile, "Error %d from pcre_fullinfo(%d)\n", rc, option); 642} 643 644 645 646/************************************************* 647* Byte flipping function * 648*************************************************/ 649 650static unsigned long int 651byteflip(unsigned long int value, int n) 652{ 653if (n == 2) return ((value & 0x00ff) << 8) | ((value & 0xff00) >> 8); 654return ((value & 0x000000ff) << 24) | 655 ((value & 0x0000ff00) << 8) | 656 ((value & 0x00ff0000) >> 8) | 657 ((value & 0xff000000) >> 24); 658} 659 660 661 662 663/************************************************* 664* Check match or recursion limit * 665*************************************************/ 666 667static int 668check_match_limit(pcre *re, pcre_extra *extra, uschar *bptr, int len, 669 int start_offset, int options, int *use_offsets, int use_size_offsets, 670 int flag, unsigned long int *limit, int errnumber, const char *msg) 671{ 672int count; 673int min = 0; 674int mid = 64; 675int max = -1; 676 677extra->flags |= flag; 678 679for (;;) 680 { 681 *limit = mid; 682 683 count = pcre_exec(re, extra, (char *)bptr, len, start_offset, options, 684 use_offsets, use_size_offsets); 685 686 if (count == errnumber) 687 { 688 /* fprintf(outfile, "Testing %s limit = %d\n", msg, mid); */ 689 min = mid; 690 mid = (mid == max - 1)? max : (max > 0)? (min + max)/2 : mid*2; 691 } 692 693 else if (count >= 0 || count == PCRE_ERROR_NOMATCH || 694 count == PCRE_ERROR_PARTIAL) 695 { 696 if (mid == min + 1) 697 { 698 fprintf(outfile, "Minimum %s limit = %d\n", msg, mid); 699 break; 700 } 701 /* fprintf(outfile, "Testing %s limit = %d\n", msg, mid); */ 702 max = mid; 703 mid = (min + mid)/2; 704 } 705 else break; /* Some other error */ 706 } 707 708extra->flags &= ~flag; 709return count; 710} 711 712 713 714/************************************************* 715* Case-independent strncmp() function * 716*************************************************/ 717 718/* 719Arguments: 720 s first string 721 t second string 722 n number of characters to compare 723 724Returns: < 0, = 0, or > 0, according to the comparison 725*/ 726 727static int 728strncmpic(uschar *s, uschar *t, int n) 729{ 730while (n--) 731 { 732 int c = tolower(*s++) - tolower(*t++); 733 if (c) return c; 734 } 735return 0; 736} 737 738 739 740/************************************************* 741* Check newline indicator * 742*************************************************/ 743 744/* This is used both at compile and run-time to check for <xxx> escapes, where 745xxx is LF, CR, CRLF, ANYCRLF, or ANY. Print a message and return 0 if there is 746no match. 747 748Arguments: 749 p points after the leading '<' 750 f file for error message 751 752Returns: appropriate PCRE_NEWLINE_xxx flags, or 0 753*/ 754 755static int 756check_newline(uschar *p, FILE *f) 757{ 758if (strncmpic(p, (uschar *)"cr>", 3) == 0) return PCRE_NEWLINE_CR; 759if (strncmpic(p, (uschar *)"lf>", 3) == 0) return PCRE_NEWLINE_LF; 760if (strncmpic(p, (uschar *)"crlf>", 5) == 0) return PCRE_NEWLINE_CRLF; 761if (strncmpic(p, (uschar *)"anycrlf>", 8) == 0) return PCRE_NEWLINE_ANYCRLF; 762if (strncmpic(p, (uschar *)"any>", 4) == 0) return PCRE_NEWLINE_ANY; 763if (strncmpic(p, (uschar *)"bsr_anycrlf>", 12) == 0) return PCRE_BSR_ANYCRLF; 764if (strncmpic(p, (uschar *)"bsr_unicode>", 12) == 0) return PCRE_BSR_UNICODE; 765fprintf(f, "Unknown newline type at: <%s\n", p); 766return 0; 767} 768 769 770 771/************************************************* 772* Usage function * 773*************************************************/ 774 775static void 776usage(void) 777{ 778printf("Usage: pcretest [options] [<input file> [<output file>]]\n\n"); 779printf("Input and output default to stdin and stdout.\n"); 780#ifdef SUPPORT_LIBREADLINE 781printf("If input is a terminal, readline() is used to read from it.\n"); 782#else 783printf("This version of pcretest is not linked with readline().\n"); 784#endif 785printf("\nOptions:\n"); 786printf(" -b show compiled code (bytecode)\n"); 787printf(" -C show PCRE compile-time options and exit\n"); 788printf(" -d debug: show compiled code and information (-b and -i)\n"); 789#if !defined NODFA 790printf(" -dfa force DFA matching for all subjects\n"); 791#endif 792printf(" -help show usage information\n"); 793printf(" -i show information about compiled patterns\n" 794 " -M find MATCH_LIMIT minimum for each subject\n" 795 " -m output memory used information\n" 796 " -o <n> set size of offsets vector to <n>\n"); 797#if !defined NOPOSIX 798printf(" -p use POSIX interface\n"); 799#endif 800printf(" -q quiet: do not output PCRE version number at start\n"); 801printf(" -S <n> set stack size to <n> megabytes\n"); 802printf(" -s output store (memory) used information\n" 803 " -t time compilation and execution\n"); 804printf(" -t <n> time compilation and execution, repeating <n> times\n"); 805printf(" -tm time execution (matching) only\n"); 806printf(" -tm <n> time execution (matching) only, repeating <n> times\n"); 807} 808 809 810 811/************************************************* 812* Main Program * 813*************************************************/ 814 815/* Read lines from named file or stdin and write to named file or stdout; lines 816consist of a regular expression, in delimiters and optionally followed by 817options, followed by a set of test data, terminated by an empty line. */ 818 819int main(int argc, char **argv) 820{ 821FILE *infile = stdin; 822int options = 0; 823int study_options = 0; 824int default_find_match_limit = FALSE; 825int op = 1; 826int timeit = 0; 827int timeitm = 0; 828int showinfo = 0; 829int showstore = 0; 830int quiet = 0; 831int size_offsets = 45; 832int size_offsets_max; 833int *offsets = NULL; 834#if !defined NOPOSIX 835int posix = 0; 836#endif 837int debug = 0; 838int done = 0; 839int all_use_dfa = 0; 840int yield = 0; 841int stack_size; 842 843/* These vectors store, end-to-end, a list of captured substring names. Assume 844that 1024 is plenty long enough for the few names we'll be testing. */ 845 846uschar copynames[1024]; 847uschar getnames[1024]; 848 849uschar *copynamesptr; 850uschar *getnamesptr; 851 852/* Get buffers from malloc() so that Electric Fence will check their misuse 853when I am debugging. They grow automatically when very long lines are read. */ 854 855buffer = (unsigned char *)malloc(buffer_size); 856dbuffer = (unsigned char *)malloc(buffer_size); 857pbuffer = (unsigned char *)malloc(buffer_size); 858 859/* The outfile variable is static so that new_malloc can use it. */ 860 861outfile = stdout; 862 863/* The following _setmode() stuff is some Windows magic that tells its runtime 864library to translate CRLF into a single LF character. At least, that's what 865I've been told: never having used Windows I take this all on trust. Originally 866it set 0x8000, but then I was advised that _O_BINARY was better. */ 867 868#if defined(_WIN32) || defined(WIN32) 869_setmode( _fileno( stdout ), _O_BINARY ); 870#endif 871 872/* Scan options */ 873 874while (argc > 1 && argv[op][0] == '-') 875 { 876 unsigned char *endptr; 877 878 if (strcmp(argv[op], "-s") == 0 || strcmp(argv[op], "-m") == 0) 879 showstore = 1; 880 else if (strcmp(argv[op], "-q") == 0) quiet = 1; 881 else if (strcmp(argv[op], "-b") == 0) debug = 1; 882 else if (strcmp(argv[op], "-i") == 0) showinfo = 1; 883 else if (strcmp(argv[op], "-d") == 0) showinfo = debug = 1; 884 else if (strcmp(argv[op], "-M") == 0) default_find_match_limit = TRUE; 885#if !defined NODFA 886 else if (strcmp(argv[op], "-dfa") == 0) all_use_dfa = 1; 887#endif 888 else if (strcmp(argv[op], "-o") == 0 && argc > 2 && 889 ((size_offsets = get_value((unsigned char *)argv[op+1], &endptr)), 890 *endptr == 0)) 891 { 892 op++; 893 argc--; 894 } 895 else if (strcmp(argv[op], "-t") == 0 || strcmp(argv[op], "-tm") == 0) 896 { 897 int both = argv[op][2] == 0; 898 int temp; 899 if (argc > 2 && (temp = get_value((unsigned char *)argv[op+1], &endptr), 900 *endptr == 0)) 901 { 902 timeitm = temp; 903 op++; 904 argc--; 905 } 906 else timeitm = LOOPREPEAT; 907 if (both) timeit = timeitm; 908 } 909 else if (strcmp(argv[op], "-S") == 0 && argc > 2 && 910 ((stack_size = get_value((unsigned char *)argv[op+1], &endptr)), 911 *endptr == 0)) 912 { 913#if defined(_WIN32) || defined(WIN32) 914 printf("PCRE: -S not supported on this OS\n"); 915 exit(1); 916#else 917 int rc; 918 struct rlimit rlim; 919 getrlimit(RLIMIT_STACK, &rlim); 920 rlim.rlim_cur = stack_size * 1024 * 1024; 921 rc = setrlimit(RLIMIT_STACK, &rlim); 922 if (rc != 0) 923 { 924 printf("PCRE: setrlimit() failed with error %d\n", rc); 925 exit(1); 926 } 927 op++; 928 argc--; 929#endif 930 } 931#if !defined NOPOSIX 932 else if (strcmp(argv[op], "-p") == 0) posix = 1; 933#endif 934 else if (strcmp(argv[op], "-C") == 0) 935 { 936 int rc; 937 unsigned long int lrc; 938 printf("PCRE version %s\n", pcre_version()); 939 printf("Compiled with\n"); 940 (void)pcre_config(PCRE_CONFIG_UTF8, &rc); 941 printf(" %sUTF-8 support\n", rc? "" : "No "); 942 (void)pcre_config(PCRE_CONFIG_UNICODE_PROPERTIES, &rc); 943 printf(" %sUnicode properties support\n", rc? "" : "No "); 944 (void)pcre_config(PCRE_CONFIG_NEWLINE, &rc); 945 /* Note that these values are always the ASCII values, even 946 in EBCDIC environments. CR is 13 and NL is 10. */ 947 printf(" Newline sequence is %s\n", (rc == 13)? "CR" : 948 (rc == 10)? "LF" : (rc == (13<<8 | 10))? "CRLF" : 949 (rc == -2)? "ANYCRLF" : 950 (rc == -1)? "ANY" : "???"); 951 (void)pcre_config(PCRE_CONFIG_BSR, &rc); 952 printf(" \\R matches %s\n", rc? "CR, LF, or CRLF only" : 953 "all Unicode newlines"); 954 (void)pcre_config(PCRE_CONFIG_LINK_SIZE, &rc); 955 printf(" Internal link size = %d\n", rc); 956 (void)pcre_config(PCRE_CONFIG_POSIX_MALLOC_THRESHOLD, &rc); 957 printf(" POSIX malloc threshold = %d\n", rc); 958 (void)pcre_config(PCRE_CONFIG_MATCH_LIMIT, &lrc); 959 printf(" Default match limit = %ld\n", lrc); 960 (void)pcre_config(PCRE_CONFIG_MATCH_LIMIT_RECURSION, &lrc); 961 printf(" Default recursion depth limit = %ld\n", lrc); 962 (void)pcre_config(PCRE_CONFIG_STACKRECURSE, &rc); 963 printf(" Match recursion uses %s\n", rc? "stack" : "heap"); 964 goto EXIT; 965 } 966 else if (strcmp(argv[op], "-help") == 0 || 967 strcmp(argv[op], "--help") == 0) 968 { 969 usage(); 970 goto EXIT; 971 } 972 else 973 { 974 printf("** Unknown or malformed option %s\n", argv[op]); 975 usage(); 976 yield = 1; 977 goto EXIT; 978 } 979 op++; 980 argc--; 981 } 982 983/* Get the store for the offsets vector, and remember what it was */ 984 985size_offsets_max = size_offsets; 986offsets = (int *)malloc(size_offsets_max * sizeof(int)); 987if (offsets == NULL) 988 { 989 printf("** Failed to get %d bytes of memory for offsets vector\n", 990 (int)(size_offsets_max * sizeof(int))); 991 yield = 1; 992 goto EXIT; 993 } 994 995/* Sort out the input and output files */ 996 997if (argc > 1) 998 { 999 infile = fopen(argv[op], INPUT_MODE); 1000 if (infile == NULL) 1001 { 1002 printf("** Failed to open %s\n", argv[op]); 1003 yield = 1; 1004 goto EXIT; 1005 } 1006 } 1007 1008if (argc > 2) 1009 { 1010 outfile = fopen(argv[op+1], OUTPUT_MODE); 1011 if (outfile == NULL) 1012 { 1013 printf("** Failed to open %s\n", argv[op+1]); 1014 yield = 1; 1015 goto EXIT; 1016 } 1017 } 1018 1019/* Set alternative malloc function */ 1020 1021pcre_malloc = new_malloc; 1022pcre_free = new_free; 1023pcre_stack_malloc = stack_malloc; 1024pcre_stack_free = stack_free; 1025 1026/* Heading line unless quiet, then prompt for first regex if stdin */ 1027 1028if (!quiet) fprintf(outfile, "PCRE version %s\n\n", pcre_version()); 1029 1030/* Main loop */ 1031 1032while (!done) 1033 { 1034 pcre *re = NULL; 1035 pcre_extra *extra = NULL; 1036 1037#if !defined NOPOSIX /* There are still compilers that require no indent */ 1038 regex_t preg; 1039 int do_posix = 0; 1040#endif 1041 1042 const char *error; 1043 unsigned char *p, *pp, *ppp; 1044 unsigned char *to_file = NULL; 1045 const unsigned char *tables = NULL; 1046 unsigned long int true_size, true_study_size = 0; 1047 size_t size, regex_gotten_store; 1048 int do_study = 0; 1049 int do_debug = debug; 1050 int do_G = 0; 1051 int do_g = 0; 1052 int do_showinfo = showinfo; 1053 int do_showrest = 0; 1054 int do_flip = 0; 1055 int erroroffset, len, delimiter, poffset; 1056 1057 use_utf8 = 0; 1058 debug_lengths = 1; 1059 1060 if (extend_inputline(infile, buffer, " re> ") == NULL) break; 1061 if (infile != stdin) fprintf(outfile, "%s", (char *)buffer); 1062 fflush(outfile); 1063 1064 p = buffer; 1065 while (isspace(*p)) p++; 1066 if (*p == 0) continue; 1067 1068 /* See if the pattern is to be loaded pre-compiled from a file. */ 1069 1070 if (*p == '<' && strchr((char *)(p+1), '<') == NULL) 1071 { 1072 unsigned long int magic, get_options; 1073 uschar sbuf[8]; 1074 FILE *f; 1075 1076 p++; 1077 pp = p + (int)strlen((char *)p); 1078 while (isspace(pp[-1])) pp--; 1079 *pp = 0; 1080 1081 f = fopen((char *)p, "rb"); 1082 if (f == NULL) 1083 { 1084 fprintf(outfile, "Failed to open %s: %s\n", p, strerror(errno)); 1085 continue; 1086 } 1087 1088 if (fread(sbuf, 1, 8, f) != 8) goto FAIL_READ; 1089 1090 true_size = 1091 (sbuf[0] << 24) | (sbuf[1] << 16) | (sbuf[2] << 8) | sbuf[3]; 1092 true_study_size = 1093 (sbuf[4] << 24) | (sbuf[5] << 16) | (sbuf[6] << 8) | sbuf[7]; 1094 1095 re = (real_pcre *)new_malloc(true_size); 1096 regex_gotten_store = gotten_store; 1097 1098 if (fread(re, 1, true_size, f) != true_size) goto FAIL_READ; 1099 1100 magic = ((real_pcre *)re)->magic_number; 1101 if (magic != MAGIC_NUMBER) 1102 { 1103 if (byteflip(magic, sizeof(magic)) == MAGIC_NUMBER) 1104 { 1105 do_flip = 1; 1106 } 1107 else 1108 { 1109 fprintf(outfile, "Data in %s is not a compiled PCRE regex\n", p); 1110 fclose(f); 1111 continue; 1112 } 1113 } 1114 1115 fprintf(outfile, "Compiled regex%s loaded from %s\n", 1116 do_flip? " (byte-inverted)" : "", p); 1117 1118 /* Need to know if UTF-8 for printing data strings */ 1119 1120 new_info(re, NULL, PCRE_INFO_OPTIONS, &get_options); 1121 use_utf8 = (get_options & PCRE_UTF8) != 0; 1122 1123 /* Now see if there is any following study data */ 1124 1125 if (true_study_size != 0) 1126 { 1127 pcre_study_data *psd; 1128 1129 extra = (pcre_extra *)new_malloc(sizeof(pcre_extra) + true_study_size); 1130 extra->flags = PCRE_EXTRA_STUDY_DATA; 1131 1132 psd = (pcre_study_data *)(((char *)extra) + sizeof(pcre_extra)); 1133 extra->study_data = psd; 1134 1135 if (fread(psd, 1, true_study_size, f) != true_study_size) 1136 { 1137 FAIL_READ: 1138 fprintf(outfile, "Failed to read data from %s\n", p); 1139 if (extra != NULL) new_free(extra); 1140 if (re != NULL) new_free(re); 1141 fclose(f); 1142 continue; 1143 } 1144 fprintf(outfile, "Study data loaded from %s\n", p); 1145 do_study = 1; /* To get the data output if requested */ 1146 } 1147 else fprintf(outfile, "No study data\n"); 1148 1149 fclose(f); 1150 goto SHOW_INFO; 1151 } 1152 1153 /* In-line pattern (the usual case). Get the delimiter and seek the end of 1154 the pattern; if is isn't complete, read more. */ 1155 1156 delimiter = *p++; 1157 1158 if (isalnum(delimiter) || delimiter == '\\') 1159 { 1160 fprintf(outfile, "** Delimiter must not be alphanumeric or \\\n"); 1161 goto SKIP_DATA; 1162 } 1163 1164 pp = p; 1165 poffset = p - buffer; 1166 1167 for(;;) 1168 { 1169 while (*pp != 0) 1170 { 1171 if (*pp == '\\' && pp[1] != 0) pp++; 1172 else if (*pp == delimiter) break; 1173 pp++; 1174 } 1175 if (*pp != 0) break; 1176 if ((pp = extend_inputline(infile, pp, " > ")) == NULL) 1177 { 1178 fprintf(outfile, "** Unexpected EOF\n"); 1179 done = 1; 1180 goto CONTINUE; 1181 } 1182 if (infile != stdin) fprintf(outfile, "%s", (char *)pp); 1183 } 1184 1185 /* The buffer may have moved while being extended; reset the start of data 1186 pointer to the correct relative point in the buffer. */ 1187 1188 p = buffer + poffset; 1189 1190 /* If the first character after the delimiter is backslash, make 1191 the pattern end with backslash. This is purely to provide a way 1192 of testing for the error message when a pattern ends with backslash. */ 1193 1194 if (pp[1] == '\\') *pp++ = '\\'; 1195 1196 /* Terminate the pattern at the delimiter, and save a copy of the pattern 1197 for callouts. */ 1198 1199 *pp++ = 0; 1200 strcpy((char *)pbuffer, (char *)p); 1201 1202 /* Look for options after final delimiter */ 1203 1204 options = 0; 1205 study_options = 0; 1206 log_store = showstore; /* default from command line */ 1207 1208 while (*pp != 0) 1209 { 1210 switch (*pp++) 1211 { 1212 case 'f': options |= PCRE_FIRSTLINE; break; 1213 case 'g': do_g = 1; break; 1214 case 'i': options |= PCRE_CASELESS; break; 1215 case 'm': options |= PCRE_MULTILINE; break; 1216 case 's': options |= PCRE_DOTALL; break; 1217 case 'x': options |= PCRE_EXTENDED; break; 1218 1219 case '+': do_showrest = 1; break; 1220 case 'A': options |= PCRE_ANCHORED; break; 1221 case 'B': do_debug = 1; break; 1222 case 'C': options |= PCRE_AUTO_CALLOUT; break; 1223 case 'D': do_debug = do_showinfo = 1; break; 1224 case 'E': options |= PCRE_DOLLAR_ENDONLY; break; 1225 case 'F': do_flip = 1; break; 1226 case 'G': do_G = 1; break; 1227 case 'I': do_showinfo = 1; break; 1228 case 'J': options |= PCRE_DUPNAMES; break; 1229 case 'M': log_store = 1; break; 1230 case 'N': options |= PCRE_NO_AUTO_CAPTURE; break; 1231 1232#if !defined NOPOSIX 1233 case 'P': do_posix = 1; break; 1234#endif 1235 1236 case 'S': do_study = 1; break; 1237 case 'U': options |= PCRE_UNGREEDY; break; 1238 case 'X': options |= PCRE_EXTRA; break; 1239 case 'Z': debug_lengths = 0; break; 1240 case '8': options |= PCRE_UTF8; use_utf8 = 1; break; 1241 case '?': options |= PCRE_NO_UTF8_CHECK; break; 1242 1243 case 'L': 1244 ppp = pp; 1245 /* The '\r' test here is so that it works on Windows. */ 1246 /* The '0' test is just in case this is an unterminated line. */ 1247 while (*ppp != 0 && *ppp != '\n' && *ppp != '\r' && *ppp != ' ') ppp++; 1248 *ppp = 0; 1249 if (setlocale(LC_CTYPE, (const char *)pp) == NULL) 1250 { 1251 fprintf(outfile, "** Failed to set locale \"%s\"\n", pp); 1252 goto SKIP_DATA; 1253 } 1254 locale_set = 1; 1255 tables = pcre_maketables(); 1256 pp = ppp; 1257 break; 1258 1259 case '>': 1260 to_file = pp; 1261 while (*pp != 0) pp++; 1262 while (isspace(pp[-1])) pp--; 1263 *pp = 0; 1264 break; 1265 1266 case '<': 1267 { 1268 if (strncmp((char *)pp, "JS>", 3) == 0) 1269 { 1270 options |= PCRE_JAVASCRIPT_COMPAT; 1271 pp += 3; 1272 } 1273 else 1274 { 1275 int x = check_newline(pp, outfile); 1276 if (x == 0) goto SKIP_DATA; 1277 options |= x; 1278 while (*pp++ != '>'); 1279 } 1280 } 1281 break; 1282 1283 case '\r': /* So that it works in Windows */ 1284 case '\n': 1285 case ' ': 1286 break; 1287 1288 default: 1289 fprintf(outfile, "** Unknown option '%c'\n", pp[-1]); 1290 goto SKIP_DATA; 1291 } 1292 } 1293 1294 /* Handle compiling via the POSIX interface, which doesn't support the 1295 timing, showing, or debugging options, nor the ability to pass over 1296 local character tables. */ 1297 1298#if !defined NOPOSIX 1299 if (posix || do_posix) 1300 { 1301 int rc; 1302 int cflags = 0; 1303 1304 if ((options & PCRE_CASELESS) != 0) cflags |= REG_ICASE; 1305 if ((options & PCRE_MULTILINE) != 0) cflags |= REG_NEWLINE; 1306 if ((options & PCRE_DOTALL) != 0) cflags |= REG_DOTALL; 1307 if ((options & PCRE_NO_AUTO_CAPTURE) != 0) cflags |= REG_NOSUB; 1308 if ((options & PCRE_UTF8) != 0) cflags |= REG_UTF8; 1309 if ((options & PCRE_UNGREEDY) != 0) cflags |= REG_UNGREEDY; 1310 1311 rc = regcomp(&preg, (char *)p, cflags); 1312 1313 /* Compilation failed; go back for another re, skipping to blank line 1314 if non-interactive. */ 1315 1316 if (rc != 0) 1317 { 1318 (void)regerror(rc, &preg, (char *)buffer, buffer_size); 1319 fprintf(outfile, "Failed: POSIX code %d: %s\n", rc, buffer); 1320 goto SKIP_DATA; 1321 } 1322 } 1323 1324 /* Handle compiling via the native interface */ 1325 1326 else 1327#endif /* !defined NOPOSIX */ 1328 1329 { 1330 unsigned long int get_options; 1331 1332 if (timeit > 0) 1333 { 1334 register int i; 1335 clock_t time_taken; 1336 clock_t start_time = clock(); 1337 for (i = 0; i < timeit; i++) 1338 { 1339 re = pcre_compile((char *)p, options, &error, &erroroffset, tables); 1340 if (re != NULL) free(re); 1341 } 1342 time_taken = clock() - start_time; 1343 fprintf(outfile, "Compile time %.4f milliseconds\n", 1344 (((double)time_taken * 1000.0) / (double)timeit) / 1345 (double)CLOCKS_PER_SEC); 1346 } 1347 1348 re = pcre_compile((char *)p, options, &error, &erroroffset, tables); 1349 1350 /* Compilation failed; go back for another re, skipping to blank line 1351 if non-interactive. */ 1352 1353 if (re == NULL) 1354 { 1355 fprintf(outfile, "Failed: %s at offset %d\n", error, erroroffset); 1356 SKIP_DATA: 1357 if (infile != stdin) 1358 { 1359 for (;;) 1360 { 1361 if (extend_inputline(infile, buffer, NULL) == NULL) 1362 { 1363 done = 1; 1364 goto CONTINUE; 1365 } 1366 len = (int)strlen((char *)buffer); 1367 while (len > 0 && isspace(buffer[len-1])) len--; 1368 if (len == 0) break; 1369 } 1370 fprintf(outfile, "\n"); 1371 } 1372 goto CONTINUE; 1373 } 1374 1375 /* Compilation succeeded. It is now possible to set the UTF-8 option from 1376 within the regex; check for this so that we know how to process the data 1377 lines. */ 1378 1379 new_info(re, NULL, PCRE_INFO_OPTIONS, &get_options); 1380 if ((get_options & PCRE_UTF8) != 0) use_utf8 = 1; 1381 1382 /* Print information if required. There are now two info-returning 1383 functions. The old one has a limited interface and returns only limited 1384 data. Check that it agrees with the newer one. */ 1385 1386 if (log_store) 1387 fprintf(outfile, "Memory allocation (code space): %d\n", 1388 (int)(gotten_store - 1389 sizeof(real_pcre) - 1390 ((real_pcre *)re)->name_count * ((real_pcre *)re)->name_entry_size)); 1391 1392 /* Extract the size for possible writing before possibly flipping it, 1393 and remember the store that was got. */ 1394 1395 true_size = ((real_pcre *)re)->size; 1396 regex_gotten_store = gotten_store; 1397 1398 /* If /S was present, study the regexp to generate additional info to 1399 help with the matching. */ 1400 1401 if (do_study) 1402 { 1403 if (timeit > 0) 1404 { 1405 register int i; 1406 clock_t time_taken; 1407 clock_t start_time = clock(); 1408 for (i = 0; i < timeit; i++) 1409 extra = pcre_study(re, study_options, &error); 1410 time_taken = clock() - start_time; 1411 if (extra != NULL) free(extra); 1412 fprintf(outfile, " Study time %.4f milliseconds\n", 1413 (((double)time_taken * 1000.0) / (double)timeit) / 1414 (double)CLOCKS_PER_SEC); 1415 } 1416 extra = pcre_study(re, study_options, &error); 1417 if (error != NULL) 1418 fprintf(outfile, "Failed to study: %s\n", error); 1419 else if (extra != NULL) 1420 true_study_size = ((pcre_study_data *)(extra->study_data))->size; 1421 } 1422 1423 /* If the 'F' option was present, we flip the bytes of all the integer 1424 fields in the regex data block and the study block. This is to make it 1425 possible to test PCRE's handling of byte-flipped patterns, e.g. those 1426 compiled on a different architecture. */ 1427 1428 if (do_flip) 1429 { 1430 real_pcre *rre = (real_pcre *)re; 1431 rre->magic_number = 1432 byteflip(rre->magic_number, sizeof(rre->magic_number)); 1433 rre->size = byteflip(rre->size, sizeof(rre->size)); 1434 rre->options = byteflip(rre->options, sizeof(rre->options)); 1435 rre->flags = (pcre_uint16)byteflip(rre->flags, sizeof(rre->flags)); 1436 rre->top_bracket = 1437 (pcre_uint16)byteflip(rre->top_bracket, sizeof(rre->top_bracket)); 1438 rre->top_backref = 1439 (pcre_uint16)byteflip(rre->top_backref, sizeof(rre->top_backref)); 1440 rre->first_byte = 1441 (pcre_uint16)byteflip(rre->first_byte, sizeof(rre->first_byte)); 1442 rre->req_byte = 1443 (pcre_uint16)byteflip(rre->req_byte, sizeof(rre->req_byte)); 1444 rre->name_table_offset = (pcre_uint16)byteflip(rre->name_table_offset, 1445 sizeof(rre->name_table_offset)); 1446 rre->name_entry_size = (pcre_uint16)byteflip(rre->name_entry_size, 1447 sizeof(rre->name_entry_size)); 1448 rre->name_count = (pcre_uint16)byteflip(rre->name_count, 1449 sizeof(rre->name_count)); 1450 1451 if (extra != NULL) 1452 { 1453 pcre_study_data *rsd = (pcre_study_data *)(extra->study_data); 1454 rsd->size = byteflip(rsd->size, sizeof(rsd->size)); 1455 rsd->flags = byteflip(rsd->flags, sizeof(rsd->flags)); 1456 rsd->minlength = byteflip(rsd->minlength, sizeof(rsd->minlength)); 1457 } 1458 } 1459 1460 /* Extract information from the compiled data if required */ 1461 1462 SHOW_INFO: 1463 1464 if (do_debug) 1465 { 1466 fprintf(outfile, "------------------------------------------------------------------\n"); 1467 pcre_printint(re, outfile, debug_lengths); 1468 } 1469 1470 /* We already have the options in get_options (see above) */ 1471 1472 if (do_showinfo) 1473 { 1474 unsigned long int all_options; 1475#if !defined NOINFOCHECK 1476 int old_first_char, old_options, old_count; 1477#endif 1478 int count, backrefmax, first_char, need_char, okpartial, jchanged, 1479 hascrorlf; 1480 int nameentrysize, namecount; 1481 const uschar *nametable; 1482 1483 new_info(re, NULL, PCRE_INFO_SIZE, &size); 1484 new_info(re, NULL, PCRE_INFO_CAPTURECOUNT, &count); 1485 new_info(re, NULL, PCRE_INFO_BACKREFMAX, &backrefmax); 1486 new_info(re, NULL, PCRE_INFO_FIRSTBYTE, &first_char); 1487 new_info(re, NULL, PCRE_INFO_LASTLITERAL, &need_char); 1488 new_info(re, NULL, PCRE_INFO_NAMEENTRYSIZE, &nameentrysize); 1489 new_info(re, NULL, PCRE_INFO_NAMECOUNT, &namecount); 1490 new_info(re, NULL, PCRE_INFO_NAMETABLE, (void *)&nametable); 1491 new_info(re, NULL, PCRE_INFO_OKPARTIAL, &okpartial); 1492 new_info(re, NULL, PCRE_INFO_JCHANGED, &jchanged); 1493 new_info(re, NULL, PCRE_INFO_HASCRORLF, &hascrorlf); 1494 1495#if !defined NOINFOCHECK 1496 old_count = pcre_info(re, &old_options, &old_first_char); 1497 if (count < 0) fprintf(outfile, 1498 "Error %d from pcre_info()\n", count); 1499 else 1500 { 1501 if (old_count != count) fprintf(outfile, 1502 "Count disagreement: pcre_fullinfo=%d pcre_info=%d\n", count, 1503 old_count); 1504 1505 if (old_first_char != first_char) fprintf(outfile, 1506 "First char disagreement: pcre_fullinfo=%d pcre_info=%d\n", 1507 first_char, old_first_char); 1508 1509 if (old_options != (int)get_options) fprintf(outfile, 1510 "Options disagreement: pcre_fullinfo=%ld pcre_info=%d\n", 1511 get_options, old_options); 1512 } 1513#endif 1514 1515 if (size != regex_gotten_store) fprintf(outfile, 1516 "Size disagreement: pcre_fullinfo=%d call to malloc for %d\n", 1517 (int)size, (int)regex_gotten_store); 1518 1519 fprintf(outfile, "Capturing subpattern count = %d\n", count); 1520 if (backrefmax > 0) 1521 fprintf(outfile, "Max back reference = %d\n", backrefmax); 1522 1523 if (namecount > 0) 1524 { 1525 fprintf(outfile, "Named capturing subpatterns:\n"); 1526 while (namecount-- > 0) 1527 { 1528 fprintf(outfile, " %s %*s%3d\n", nametable + 2, 1529 nameentrysize - 3 - (int)strlen((char *)nametable + 2), "", 1530 GET2(nametable, 0)); 1531 nametable += nameentrysize; 1532 } 1533 } 1534 1535 if (!okpartial) fprintf(outfile, "Partial matching not supported\n"); 1536 if (hascrorlf) fprintf(outfile, "Contains explicit CR or LF match\n"); 1537 1538 all_options = ((real_pcre *)re)->options; 1539 if (do_flip) all_options = byteflip(all_options, sizeof(all_options)); 1540 1541 if (get_options == 0) fprintf(outfile, "No options\n"); 1542 else fprintf(outfile, "Options:%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s\n", 1543 ((get_options & PCRE_ANCHORED) != 0)? " anchored" : "", 1544 ((get_options & PCRE_CASELESS) != 0)? " caseless" : "", 1545 ((get_options & PCRE_EXTENDED) != 0)? " extended" : "", 1546 ((get_options & PCRE_MULTILINE) != 0)? " multiline" : "", 1547 ((get_options & PCRE_FIRSTLINE) != 0)? " firstline" : "", 1548 ((get_options & PCRE_DOTALL) != 0)? " dotall" : "", 1549 ((get_options & PCRE_BSR_ANYCRLF) != 0)? " bsr_anycrlf" : "", 1550 ((get_options & PCRE_BSR_UNICODE) != 0)? " bsr_unicode" : "", 1551 ((get_options & PCRE_DOLLAR_ENDONLY) != 0)? " dollar_endonly" : "", 1552 ((get_options & PCRE_EXTRA) != 0)? " extra" : "", 1553 ((get_options & PCRE_UNGREEDY) != 0)? " ungreedy" : "", 1554 ((get_options & PCRE_NO_AUTO_CAPTURE) != 0)? " no_auto_capture" : "", 1555 ((get_options & PCRE_UTF8) != 0)? " utf8" : "", 1556 ((get_options & PCRE_NO_UTF8_CHECK) != 0)? " no_utf8_check" : "", 1557 ((get_options & PCRE_DUPNAMES) != 0)? " dupnames" : ""); 1558 1559 if (jchanged) fprintf(outfile, "Duplicate name status changes\n"); 1560 1561 switch (get_options & PCRE_NEWLINE_BITS) 1562 { 1563 case PCRE_NEWLINE_CR: 1564 fprintf(outfile, "Forced newline sequence: CR\n"); 1565 break; 1566 1567 case PCRE_NEWLINE_LF: 1568 fprintf(outfile, "Forced newline sequence: LF\n"); 1569 break; 1570 1571 case PCRE_NEWLINE_CRLF: 1572 fprintf(outfile, "Forced newline sequence: CRLF\n"); 1573 break; 1574 1575 case PCRE_NEWLINE_ANYCRLF: 1576 fprintf(outfile, "Forced newline sequence: ANYCRLF\n"); 1577 break; 1578 1579 case PCRE_NEWLINE_ANY: 1580 fprintf(outfile, "Forced newline sequence: ANY\n"); 1581 break; 1582 1583 default: 1584 break; 1585 } 1586 1587 if (first_char == -1) 1588 { 1589 fprintf(outfile, "First char at start or follows newline\n"); 1590 } 1591 else if (first_char < 0) 1592 { 1593 fprintf(outfile, "No first char\n"); 1594 } 1595 else 1596 { 1597 int ch = first_char & 255; 1598 const char *caseless = ((first_char & REQ_CASELESS) == 0)? 1599 "" : " (caseless)"; 1600 if (PRINTHEX(ch)) 1601 fprintf(outfile, "First char = \'%c\'%s\n", ch, caseless); 1602 else 1603 fprintf(outfile, "First char = %d%s\n", ch, caseless); 1604 } 1605 1606 if (need_char < 0) 1607 { 1608 fprintf(outfile, "No need char\n"); 1609 } 1610 else 1611 { 1612 int ch = need_char & 255; 1613 const char *caseless = ((need_char & REQ_CASELESS) == 0)? 1614 "" : " (caseless)"; 1615 if (PRINTHEX(ch)) 1616 fprintf(outfile, "Need char = \'%c\'%s\n", ch, caseless); 1617 else 1618 fprintf(outfile, "Need char = %d%s\n", ch, caseless); 1619 } 1620 1621 /* Don't output study size; at present it is in any case a fixed 1622 value, but it varies, depending on the computer architecture, and 1623 so messes up the test suite. (And with the /F option, it might be 1624 flipped.) */ 1625 1626 if (do_study) 1627 { 1628 if (extra == NULL) 1629 fprintf(outfile, "Study returned NULL\n"); 1630 else 1631 { 1632 uschar *start_bits = NULL; 1633 int minlength; 1634 1635 new_info(re, extra, PCRE_INFO_MINLENGTH, &minlength); 1636 fprintf(outfile, "Subject length lower bound = %d\n", minlength); 1637 1638 new_info(re, extra, PCRE_INFO_FIRSTTABLE, &start_bits); 1639 if (start_bits == NULL) 1640 fprintf(outfile, "No set of starting bytes\n"); 1641 else 1642 { 1643 int i; 1644 int c = 24; 1645 fprintf(outfile, "Starting byte set: "); 1646 for (i = 0; i < 256; i++) 1647 { 1648 if ((start_bits[i/8] & (1<<(i&7))) != 0) 1649 { 1650 if (c > 75) 1651 { 1652 fprintf(outfile, "\n "); 1653 c = 2; 1654 } 1655 if (PRINTHEX(i) && i != ' ') 1656 { 1657 fprintf(outfile, "%c ", i); 1658 c += 2; 1659 } 1660 else 1661 { 1662 fprintf(outfile, "\\x%02x ", i); 1663 c += 5; 1664 } 1665 } 1666 } 1667 fprintf(outfile, "\n"); 1668 } 1669 } 1670 } 1671 } 1672 1673 /* If the '>' option was present, we write out the regex to a file, and 1674 that is all. The first 8 bytes of the file are the regex length and then 1675 the study length, in big-endian order. */ 1676 1677 if (to_file != NULL) 1678 { 1679 FILE *f = fopen((char *)to_file, "wb"); 1680 if (f == NULL) 1681 { 1682 fprintf(outfile, "Unable to open %s: %s\n", to_file, strerror(errno)); 1683 } 1684 else 1685 { 1686 uschar sbuf[8]; 1687 sbuf[0] = (uschar)((true_size >> 24) & 255); 1688 sbuf[1] = (uschar)((true_size >> 16) & 255); 1689 sbuf[2] = (uschar)((true_size >> 8) & 255); 1690 sbuf[3] = (uschar)((true_size) & 255); 1691 1692 sbuf[4] = (uschar)((true_study_size >> 24) & 255); 1693 sbuf[5] = (uschar)((true_study_size >> 16) & 255); 1694 sbuf[6] = (uschar)((true_study_size >> 8) & 255); 1695 sbuf[7] = (uschar)((true_study_size) & 255); 1696 1697 if (fwrite(sbuf, 1, 8, f) < 8 || 1698 fwrite(re, 1, true_size, f) < true_size) 1699 { 1700 fprintf(outfile, "Write error on %s: %s\n", to_file, strerror(errno)); 1701 } 1702 else 1703 { 1704 fprintf(outfile, "Compiled regex written to %s\n", to_file); 1705 if (extra != NULL) 1706 { 1707 if (fwrite(extra->study_data, 1, true_study_size, f) < 1708 true_study_size) 1709 { 1710 fprintf(outfile, "Write error on %s: %s\n", to_file, 1711 strerror(errno)); 1712 } 1713 else fprintf(outfile, "Study data written to %s\n", to_file); 1714 1715 } 1716 } 1717 fclose(f); 1718 } 1719 1720 new_free(re); 1721 if (extra != NULL) new_free(extra); 1722 if (tables != NULL) new_free((void *)tables); 1723 continue; /* With next regex */ 1724 } 1725 } /* End of non-POSIX compile */ 1726 1727 /* Read data lines and test them */ 1728 1729 for (;;) 1730 { 1731 uschar *q; 1732 uschar *bptr; 1733 int *use_offsets = offsets; 1734 int use_size_offsets = size_offsets; 1735 int callout_data = 0; 1736 int callout_data_set = 0; 1737 int count, c; 1738 int copystrings = 0; 1739 int find_match_limit = default_find_match_limit; 1740 int getstrings = 0; 1741 int getlist = 0; 1742 int gmatched = 0; 1743 int start_offset = 0; 1744 int g_notempty = 0; 1745 int use_dfa = 0; 1746 1747 options = 0; 1748 1749 *copynames = 0; 1750 *getnames = 0; 1751 1752 copynamesptr = copynames; 1753 getnamesptr = getnames; 1754 1755 pcre_callout = callout; 1756 first_callout = 1; 1757 callout_extra = 0; 1758 callout_count = 0; 1759 callout_fail_count = 999999; 1760 callout_fail_id = -1; 1761 show_malloc = 0; 1762 1763 if (extra != NULL) extra->flags &= 1764 ~(PCRE_EXTRA_MATCH_LIMIT|PCRE_EXTRA_MATCH_LIMIT_RECURSION); 1765 1766 len = 0; 1767 for (;;) 1768 { 1769 if (extend_inputline(infile, buffer + len, "data> ") == NULL) 1770 { 1771 if (len > 0) break; 1772 done = 1; 1773 goto CONTINUE; 1774 } 1775 if (infile != stdin) fprintf(outfile, "%s", (char *)buffer); 1776 len = (int)strlen((char *)buffer); 1777 if (buffer[len-1] == '\n') break; 1778 } 1779 1780 while (len > 0 && isspace(buffer[len-1])) len--; 1781 buffer[len] = 0; 1782 if (len == 0) break; 1783 1784 p = buffer; 1785 while (isspace(*p)) p++; 1786 1787 bptr = q = dbuffer; 1788 while ((c = *p++) != 0) 1789 { 1790 int i = 0; 1791 int n = 0; 1792 1793 if (c == '\\') switch ((c = *p++)) 1794 { 1795 case 'a': c = 7; break; 1796 case 'b': c = '\b'; break; 1797 case 'e': c = 27; break; 1798 case 'f': c = '\f'; break; 1799 case 'n': c = '\n'; break; 1800 case 'r': c = '\r'; break; 1801 case 't': c = '\t'; break; 1802 case 'v': c = '\v'; break; 1803 1804 case '0': case '1': case '2': case '3': 1805 case '4': case '5': case '6': case '7': 1806 c -= '0'; 1807 while (i++ < 2 && isdigit(*p) && *p != '8' && *p != '9') 1808 c = c * 8 + *p++ - '0'; 1809 1810#if !defined NOUTF8 1811 if (use_utf8 && c > 255) 1812 { 1813 unsigned char buff8[8]; 1814 int ii, utn; 1815 utn = ord2utf8(c, buff8); 1816 for (ii = 0; ii < utn - 1; ii++) *q++ = buff8[ii]; 1817 c = buff8[ii]; /* Last byte */ 1818 } 1819#endif 1820 break; 1821 1822 case 'x': 1823 1824 /* Handle \x{..} specially - new Perl thing for utf8 */ 1825 1826#if !defined NOUTF8 1827 if (*p == '{') 1828 { 1829 unsigned char *pt = p; 1830 c = 0; 1831 while (isxdigit(*(++pt))) 1832 c = c * 16 + tolower(*pt) - ((isdigit(*pt))? '0' : 'W'); 1833 if (*pt == '}') 1834 { 1835 unsigned char buff8[8]; 1836 int ii, utn; 1837 if (use_utf8) 1838 { 1839 utn = ord2utf8(c, buff8); 1840 for (ii = 0; ii < utn - 1; ii++) *q++ = buff8[ii]; 1841 c = buff8[ii]; /* Last byte */ 1842 } 1843 else 1844 { 1845 if (c > 255) 1846 fprintf(outfile, "** Character \\x{%x} is greater than 255 and " 1847 "UTF-8 mode is not enabled.\n" 1848 "** Truncation will probably give the wrong result.\n", c); 1849 } 1850 p = pt + 1; 1851 break; 1852 } 1853 /* Not correct form; fall through */ 1854 } 1855#endif 1856 1857 /* Ordinary \x */ 1858 1859 c = 0; 1860 while (i++ < 2 && isxdigit(*p)) 1861 { 1862 c = c * 16 + tolower(*p) - ((isdigit(*p))? '0' : 'W'); 1863 p++; 1864 } 1865 break; 1866 1867 case 0: /* \ followed by EOF allows for an empty line */ 1868 p--; 1869 continue; 1870 1871 case '>': 1872 while(isdigit(*p)) start_offset = start_offset * 10 + *p++ - '0'; 1873 continue; 1874 1875 case 'A': /* Option setting */ 1876 options |= PCRE_ANCHORED; 1877 continue; 1878 1879 case 'B': 1880 options |= PCRE_NOTBOL; 1881 continue; 1882 1883 case 'C': 1884 if (isdigit(*p)) /* Set copy string */ 1885 { 1886 while(isdigit(*p)) n = n * 10 + *p++ - '0'; 1887 copystrings |= 1 << n; 1888 } 1889 else if (isalnum(*p)) 1890 { 1891 uschar *npp = copynamesptr; 1892 while (isalnum(*p)) *npp++ = *p++; 1893 *npp++ = 0; 1894 *npp = 0; 1895 n = pcre_get_stringnumber(re, (char *)copynamesptr); 1896 if (n < 0) 1897 fprintf(outfile, "no parentheses with name \"%s\"\n", copynamesptr); 1898 copynamesptr = npp; 1899 } 1900 else if (*p == '+') 1901 { 1902 callout_extra = 1; 1903 p++; 1904 } 1905 else if (*p == '-') 1906 { 1907 pcre_callout = NULL; 1908 p++; 1909 } 1910 else if (*p == '!') 1911 { 1912 callout_fail_id = 0; 1913 p++; 1914 while(isdigit(*p)) 1915 callout_fail_id = callout_fail_id * 10 + *p++ - '0'; 1916 callout_fail_count = 0; 1917 if (*p == '!') 1918 { 1919 p++; 1920 while(isdigit(*p)) 1921 callout_fail_count = callout_fail_count * 10 + *p++ - '0'; 1922 } 1923 } 1924 else if (*p == '*') 1925 { 1926 int sign = 1; 1927 callout_data = 0; 1928 if (*(++p) == '-') { sign = -1; p++; } 1929 while(isdigit(*p)) 1930 callout_data = callout_data * 10 + *p++ - '0'; 1931 callout_data *= sign; 1932 callout_data_set = 1; 1933 } 1934 continue; 1935 1936#if !defined NODFA 1937 case 'D': 1938#if !defined NOPOSIX 1939 if (posix || do_posix) 1940 printf("** Can't use dfa matching in POSIX mode: \\D ignored\n"); 1941 else 1942#endif 1943 use_dfa = 1; 1944 continue; 1945 1946 case 'F': 1947 options |= PCRE_DFA_SHORTEST; 1948 continue; 1949#endif 1950 1951 case 'G': 1952 if (isdigit(*p)) 1953 { 1954 while(isdigit(*p)) n = n * 10 + *p++ - '0'; 1955 getstrings |= 1 << n; 1956 } 1957 else if (isalnum(*p)) 1958 { 1959 uschar *npp = getnamesptr; 1960 while (isalnum(*p)) *npp++ = *p++; 1961 *npp++ = 0; 1962 *npp = 0; 1963 n = pcre_get_stringnumber(re, (char *)getnamesptr); 1964 if (n < 0) 1965 fprintf(outfile, "no parentheses with name \"%s\"\n", getnamesptr); 1966 getnamesptr = npp; 1967 } 1968 continue; 1969 1970 case 'L': 1971 getlist = 1; 1972 continue; 1973 1974 case 'M': 1975 find_match_limit = 1; 1976 continue; 1977 1978 case 'N': 1979 if ((options & PCRE_NOTEMPTY) != 0) 1980 options = (options & ~PCRE_NOTEMPTY) | PCRE_NOTEMPTY_ATSTART; 1981 else 1982 options |= PCRE_NOTEMPTY; 1983 continue; 1984 1985 case 'O': 1986 while(isdigit(*p)) n = n * 10 + *p++ - '0'; 1987 if (n > size_offsets_max) 1988 { 1989 size_offsets_max = n; 1990 free(offsets); 1991 use_offsets = offsets = (int *)malloc(size_offsets_max * sizeof(int)); 1992 if (offsets == NULL) 1993 { 1994 printf("** Failed to get %d bytes of memory for offsets vector\n", 1995 (int)(size_offsets_max * sizeof(int))); 1996 yield = 1; 1997 goto EXIT; 1998 } 1999 } 2000 use_size_offsets = n; 2001 if (n == 0) use_offsets = NULL; /* Ensures it can't write to it */ 2002 continue; 2003 2004 case 'P': 2005 options |= ((options & PCRE_PARTIAL_SOFT) == 0)? 2006 PCRE_PARTIAL_SOFT : PCRE_PARTIAL_HARD; 2007 continue; 2008 2009 case 'Q': 2010 while(isdigit(*p)) n = n * 10 + *p++ - '0'; 2011 if (extra == NULL) 2012 { 2013 extra = (pcre_extra *)malloc(sizeof(pcre_extra)); 2014 extra->flags = 0; 2015 } 2016 extra->flags |= PCRE_EXTRA_MATCH_LIMIT_RECURSION; 2017 extra->match_limit_recursion = n; 2018 continue; 2019 2020 case 'q': 2021 while(isdigit(*p)) n = n * 10 + *p++ - '0'; 2022 if (extra == NULL) 2023 { 2024 extra = (pcre_extra *)malloc(sizeof(pcre_extra)); 2025 extra->flags = 0; 2026 } 2027 extra->flags |= PCRE_EXTRA_MATCH_LIMIT; 2028 extra->match_limit = n; 2029 continue; 2030 2031#if !defined NODFA 2032 case 'R': 2033 options |= PCRE_DFA_RESTART; 2034 continue; 2035#endif 2036 2037 case 'S': 2038 show_malloc = 1; 2039 continue; 2040 2041 case 'Y': 2042 options |= PCRE_NO_START_OPTIMIZE; 2043 continue; 2044 2045 case 'Z': 2046 options |= PCRE_NOTEOL; 2047 continue; 2048 2049 case '?': 2050 options |= PCRE_NO_UTF8_CHECK; 2051 continue; 2052 2053 case '<': 2054 { 2055 int x = check_newline(p, outfile); 2056 if (x == 0) goto NEXT_DATA; 2057 options |= x; 2058 while (*p++ != '>'); 2059 } 2060 continue; 2061 } 2062 *q++ = c; 2063 } 2064 *q = 0; 2065 len = q - dbuffer; 2066 2067 /* Move the data to the end of the buffer so that a read over the end of 2068 the buffer will be seen by valgrind, even if it doesn't cause a crash. If 2069 we are using the POSIX interface, we must include the terminating zero. */ 2070 2071#if !defined NOPOSIX 2072 if (posix || do_posix) 2073 { 2074 memmove(bptr + buffer_size - len - 1, bptr, len + 1); 2075 bptr += buffer_size - len - 1; 2076 } 2077 else 2078#endif 2079 { 2080 memmove(bptr + buffer_size - len, bptr, len); 2081 bptr += buffer_size - len; 2082 } 2083 2084 if ((all_use_dfa || use_dfa) && find_match_limit) 2085 { 2086 printf("**Match limit not relevant for DFA matching: ignored\n"); 2087 find_match_limit = 0; 2088 } 2089 2090 /* Handle matching via the POSIX interface, which does not 2091 support timing or playing with the match limit or callout data. */ 2092 2093#if !defined NOPOSIX 2094 if (posix || do_posix) 2095 { 2096 int rc; 2097 int eflags = 0; 2098 regmatch_t *pmatch = NULL; 2099 if (use_size_offsets > 0) 2100 pmatch = (regmatch_t *)malloc(sizeof(regmatch_t) * use_size_offsets); 2101 if ((options & PCRE_NOTBOL) != 0) eflags |= REG_NOTBOL; 2102 if ((options & PCRE_NOTEOL) != 0) eflags |= REG_NOTEOL; 2103 if ((options & PCRE_NOTEMPTY) != 0) eflags |= REG_NOTEMPTY; 2104 2105 rc = regexec(&preg, (const char *)bptr, use_size_offsets, pmatch, eflags); 2106 2107 if (rc != 0) 2108 { 2109 (void)regerror(rc, &preg, (char *)buffer, buffer_size); 2110 fprintf(outfile, "No match: POSIX code %d: %s\n", rc, buffer); 2111 } 2112 else if ((((const pcre *)preg.re_pcre)->options & PCRE_NO_AUTO_CAPTURE) 2113 != 0) 2114 { 2115 fprintf(outfile, "Matched with REG_NOSUB\n"); 2116 } 2117 else 2118 { 2119 size_t i; 2120 for (i = 0; i < (size_t)use_size_offsets; i++) 2121 { 2122 if (pmatch[i].rm_so >= 0) 2123 { 2124 fprintf(outfile, "%2d: ", (int)i); 2125 (void)pchars(dbuffer + pmatch[i].rm_so, 2126 pmatch[i].rm_eo - pmatch[i].rm_so, outfile); 2127 fprintf(outfile, "\n"); 2128 if (i == 0 && do_showrest) 2129 { 2130 fprintf(outfile, " 0+ "); 2131 (void)pchars(dbuffer + pmatch[i].rm_eo, len - pmatch[i].rm_eo, 2132 outfile); 2133 fprintf(outfile, "\n"); 2134 } 2135 } 2136 } 2137 } 2138 free(pmatch); 2139 } 2140 2141 /* Handle matching via the native interface - repeats for /g and /G */ 2142 2143 else 2144#endif /* !defined NOPOSIX */ 2145 2146 for (;; gmatched++) /* Loop for /g or /G */ 2147 { 2148 if (timeitm > 0) 2149 { 2150 register int i; 2151 clock_t time_taken; 2152 clock_t start_time = clock(); 2153 2154#if !defined NODFA 2155 if (all_use_dfa || use_dfa) 2156 { 2157 int workspace[1000]; 2158 for (i = 0; i < timeitm; i++) 2159 count = pcre_dfa_exec(re, extra, (char *)bptr, len, start_offset, 2160 options | g_notempty, use_offsets, use_size_offsets, workspace, 2161 sizeof(workspace)/sizeof(int)); 2162 } 2163 else 2164#endif 2165 2166 for (i = 0; i < timeitm; i++) 2167 count = pcre_exec(re, extra, (char *)bptr, len, 2168 start_offset, options | g_notempty, use_offsets, use_size_offsets); 2169 2170 time_taken = clock() - start_time; 2171 fprintf(outfile, "Execute time %.4f milliseconds\n", 2172 (((double)time_taken * 1000.0) / (double)timeitm) / 2173 (double)CLOCKS_PER_SEC); 2174 } 2175 2176 /* If find_match_limit is set, we want to do repeated matches with 2177 varying limits in order to find the minimum value for the match limit and 2178 for the recursion limit. */ 2179 2180 if (find_match_limit) 2181 { 2182 if (extra == NULL) 2183 { 2184 extra = (pcre_extra *)malloc(sizeof(pcre_extra)); 2185 extra->flags = 0; 2186 } 2187 2188 (void)check_match_limit(re, extra, bptr, len, start_offset, 2189 options|g_notempty, use_offsets, use_size_offsets, 2190 PCRE_EXTRA_MATCH_LIMIT, &(extra->match_limit), 2191 PCRE_ERROR_MATCHLIMIT, "match()"); 2192 2193 count = check_match_limit(re, extra, bptr, len, start_offset, 2194 options|g_notempty, use_offsets, use_size_offsets, 2195 PCRE_EXTRA_MATCH_LIMIT_RECURSION, &(extra->match_limit_recursion), 2196 PCRE_ERROR_RECURSIONLIMIT, "match() recursion"); 2197 } 2198 2199 /* If callout_data is set, use the interface with additional data */ 2200 2201 else if (callout_data_set) 2202 { 2203 if (extra == NULL) 2204 { 2205 extra = (pcre_extra *)malloc(sizeof(pcre_extra)); 2206 extra->flags = 0; 2207 } 2208 extra->flags |= PCRE_EXTRA_CALLOUT_DATA; 2209 extra->callout_data = &callout_data; 2210 count = pcre_exec(re, extra, (char *)bptr, len, start_offset, 2211 options | g_notempty, use_offsets, use_size_offsets); 2212 extra->flags &= ~PCRE_EXTRA_CALLOUT_DATA; 2213 } 2214 2215 /* The normal case is just to do the match once, with the default 2216 value of match_limit. */ 2217 2218#if !defined NODFA 2219 else if (all_use_dfa || use_dfa) 2220 { 2221 int workspace[1000]; 2222 count = pcre_dfa_exec(re, extra, (char *)bptr, len, start_offset, 2223 options | g_notempty, use_offsets, use_size_offsets, workspace, 2224 sizeof(workspace)/sizeof(int)); 2225 if (count == 0) 2226 { 2227 fprintf(outfile, "Matched, but too many subsidiary matches\n"); 2228 count = use_size_offsets/2; 2229 } 2230 } 2231#endif 2232 2233 else 2234 { 2235 count = pcre_exec(re, extra, (char *)bptr, len, 2236 start_offset, options | g_notempty, use_offsets, use_size_offsets); 2237 if (count == 0) 2238 { 2239 fprintf(outfile, "Matched, but too many substrings\n"); 2240 count = use_size_offsets/3; 2241 } 2242 } 2243 2244 /* Matched */ 2245 2246 if (count >= 0) 2247 { 2248 int i, maxcount; 2249 2250#if !defined NODFA 2251 if (all_use_dfa || use_dfa) maxcount = use_size_offsets/2; else 2252#endif 2253 maxcount = use_size_offsets/3; 2254 2255 /* This is a check against a lunatic return value. */ 2256 2257 if (count > maxcount) 2258 { 2259 fprintf(outfile, 2260 "** PCRE error: returned count %d is too big for offset size %d\n", 2261 count, use_size_offsets); 2262 count = use_size_offsets/3; 2263 if (do_g || do_G) 2264 { 2265 fprintf(outfile, "** /%c loop abandoned\n", do_g? 'g' : 'G'); 2266 do_g = do_G = FALSE; /* Break g/G loop */ 2267 } 2268 } 2269 2270 for (i = 0; i < count * 2; i += 2) 2271 { 2272 if (use_offsets[i] < 0) 2273 fprintf(outfile, "%2d: <unset>\n", i/2); 2274 else 2275 { 2276 fprintf(outfile, "%2d: ", i/2); 2277 (void)pchars(bptr + use_offsets[i], 2278 use_offsets[i+1] - use_offsets[i], outfile); 2279 fprintf(outfile, "\n"); 2280 if (i == 0) 2281 { 2282 if (do_showrest) 2283 { 2284 fprintf(outfile, " 0+ "); 2285 (void)pchars(bptr + use_offsets[i+1], len - use_offsets[i+1], 2286 outfile); 2287 fprintf(outfile, "\n"); 2288 } 2289 } 2290 } 2291 } 2292 2293 for (i = 0; i < 32; i++) 2294 { 2295 if ((copystrings & (1 << i)) != 0) 2296 { 2297 char copybuffer[256]; 2298 int rc = pcre_copy_substring((char *)bptr, use_offsets, count, 2299 i, copybuffer, sizeof(copybuffer)); 2300 if (rc < 0) 2301 fprintf(outfile, "copy substring %d failed %d\n", i, rc); 2302 else 2303 fprintf(outfile, "%2dC %s (%d)\n", i, copybuffer, rc); 2304 } 2305 } 2306 2307 for (copynamesptr = copynames; 2308 *copynamesptr != 0; 2309 copynamesptr += (int)strlen((char*)copynamesptr) + 1) 2310 { 2311 char copybuffer[256]; 2312 int rc = pcre_copy_named_substring(re, (char *)bptr, use_offsets, 2313 count, (char *)copynamesptr, copybuffer, sizeof(copybuffer)); 2314 if (rc < 0) 2315 fprintf(outfile, "copy substring %s failed %d\n", copynamesptr, rc); 2316 else 2317 fprintf(outfile, " C %s (%d) %s\n", copybuffer, rc, copynamesptr); 2318 } 2319 2320 for (i = 0; i < 32; i++) 2321 { 2322 if ((getstrings & (1 << i)) != 0) 2323 { 2324 const char *substring; 2325 int rc = pcre_get_substring((char *)bptr, use_offsets, count, 2326 i, &substring); 2327 if (rc < 0) 2328 fprintf(outfile, "get substring %d failed %d\n", i, rc); 2329 else 2330 { 2331 fprintf(outfile, "%2dG %s (%d)\n", i, substring, rc); 2332 pcre_free_substring(substring); 2333 } 2334 } 2335 } 2336 2337 for (getnamesptr = getnames; 2338 *getnamesptr != 0; 2339 getnamesptr += (int)strlen((char*)getnamesptr) + 1) 2340 { 2341 const char *substring; 2342 int rc = pcre_get_named_substring(re, (char *)bptr, use_offsets, 2343 count, (char *)getnamesptr, &substring); 2344 if (rc < 0) 2345 fprintf(outfile, "copy substring %s failed %d\n", getnamesptr, rc); 2346 else 2347 { 2348 fprintf(outfile, " G %s (%d) %s\n", substring, rc, getnamesptr); 2349 pcre_free_substring(substring); 2350 } 2351 } 2352 2353 if (getlist) 2354 { 2355 const char **stringlist; 2356 int rc = pcre_get_substring_list((char *)bptr, use_offsets, count, 2357 &stringlist); 2358 if (rc < 0) 2359 fprintf(outfile, "get substring list failed %d\n", rc); 2360 else 2361 { 2362 for (i = 0; i < count; i++) 2363 fprintf(outfile, "%2dL %s\n", i, stringlist[i]); 2364 if (stringlist[i] != NULL) 2365 fprintf(outfile, "string list not terminated by NULL\n"); 2366 /* free((void *)stringlist); */ 2367 pcre_free_substring_list(stringlist); 2368 } 2369 } 2370 } 2371 2372 /* There was a partial match */ 2373 2374 else if (count == PCRE_ERROR_PARTIAL) 2375 { 2376 fprintf(outfile, "Partial match"); 2377 if (use_size_offsets > 1) 2378 { 2379 fprintf(outfile, ": "); 2380 pchars(bptr + use_offsets[0], use_offsets[1] - use_offsets[0], 2381 outfile); 2382 } 2383 fprintf(outfile, "\n"); 2384 break; /* Out of the /g loop */ 2385 } 2386 2387 /* Failed to match. If this is a /g or /G loop and we previously set 2388 g_notempty after a null match, this is not necessarily the end. We want 2389 to advance the start offset, and continue. We won't be at the end of the 2390 string - that was checked before setting g_notempty. 2391 2392 Complication arises in the case when the newline option is "any" or 2393 "anycrlf". If the previous match was at the end of a line terminated by 2394 CRLF, an advance of one character just passes the \r, whereas we should 2395 prefer the longer newline sequence, as does the code in pcre_exec(). 2396 Fudge the offset value to achieve this. 2397 2398 Otherwise, in the case of UTF-8 matching, the advance must be one 2399 character, not one byte. */ 2400 2401 else 2402 { 2403 if (g_notempty != 0) 2404 { 2405 int onechar = 1; 2406 unsigned int obits = ((real_pcre *)re)->options; 2407 use_offsets[0] = start_offset; 2408 if ((obits & PCRE_NEWLINE_BITS) == 0) 2409 { 2410 int d; 2411 (void)pcre_config(PCRE_CONFIG_NEWLINE, &d); 2412 /* Note that these values are always the ASCII ones, even in 2413 EBCDIC environments. CR = 13, NL = 10. */ 2414 obits = (d == 13)? PCRE_NEWLINE_CR : 2415 (d == 10)? PCRE_NEWLINE_LF : 2416 (d == (13<<8 | 10))? PCRE_NEWLINE_CRLF : 2417 (d == -2)? PCRE_NEWLINE_ANYCRLF : 2418 (d == -1)? PCRE_NEWLINE_ANY : 0; 2419 } 2420 if (((obits & PCRE_NEWLINE_BITS) == PCRE_NEWLINE_ANY || 2421 (obits & PCRE_NEWLINE_BITS) == PCRE_NEWLINE_ANYCRLF) 2422 && 2423 start_offset < len - 1 && 2424 bptr[start_offset] == '\r' && 2425 bptr[start_offset+1] == '\n') 2426 onechar++; 2427 else if (use_utf8) 2428 { 2429 while (start_offset + onechar < len) 2430 { 2431 int tb = bptr[start_offset+onechar]; 2432 if (tb <= 127) break; 2433 tb &= 0xc0; 2434 if (tb != 0 && tb != 0xc0) onechar++; 2435 } 2436 } 2437 use_offsets[1] = start_offset + onechar; 2438 } 2439 else 2440 { 2441 if (count == PCRE_ERROR_NOMATCH) 2442 { 2443 if (gmatched == 0) fprintf(outfile, "No match\n"); 2444 } 2445 else fprintf(outfile, "Error %d\n", count); 2446 break; /* Out of the /g loop */ 2447 } 2448 } 2449 2450 /* If not /g or /G we are done */ 2451 2452 if (!do_g && !do_G) break; 2453 2454 /* If we have matched an empty string, first check to see if we are at 2455 the end of the subject. If so, the /g loop is over. Otherwise, mimic what 2456 Perl's /g options does. This turns out to be rather cunning. First we set 2457 PCRE_NOTEMPTY_ATSTART and PCRE_ANCHORED and try the match again at the 2458 same point. If this fails (picked up above) we advance to the next 2459 character. */ 2460 2461 g_notempty = 0; 2462 2463 if (use_offsets[0] == use_offsets[1]) 2464 { 2465 if (use_offsets[0] == len) break; 2466 g_notempty = PCRE_NOTEMPTY_ATSTART | PCRE_ANCHORED; 2467 } 2468 2469 /* For /g, update the start offset, leaving the rest alone */ 2470 2471 if (do_g) start_offset = use_offsets[1]; 2472 2473 /* For /G, update the pointer and length */ 2474 2475 else 2476 { 2477 bptr += use_offsets[1]; 2478 len -= use_offsets[1]; 2479 } 2480 } /* End of loop for /g and /G */ 2481 2482 NEXT_DATA: continue; 2483 } /* End of loop for data lines */ 2484 2485 CONTINUE: 2486 2487#if !defined NOPOSIX 2488 if (posix || do_posix) regfree(&preg); 2489#endif 2490 2491 if (re != NULL) new_free(re); 2492 if (extra != NULL) new_free(extra); 2493 if (tables != NULL) 2494 { 2495 new_free((void *)tables); 2496 setlocale(LC_CTYPE, "C"); 2497 locale_set = 0; 2498 } 2499 } 2500 2501if (infile == stdin) fprintf(outfile, "\n"); 2502 2503EXIT: 2504 2505if (infile != NULL && infile != stdin) fclose(infile); 2506if (outfile != NULL && outfile != stdout) fclose(outfile); 2507 2508free(buffer); 2509free(dbuffer); 2510free(pbuffer); 2511free(offsets); 2512 2513return yield; 2514} 2515 2516/* End of pcretest.c */ 2517